eFinder

eFinder

Home › Technological Innovation vs. Imitation › Article

China is losing LLM race but it can still win in AI, ex-Tencent AI lead says

South China Morning Post · · 262 words · By Vincent Chow
Technological Innovation vs. Imitation US-China AI Competition
open_in_new Read the original article: https://www.scmp.com/tech/big-tech/article/3354394/china-losing-llm-race-it-can-…

psychologyDetected Techniques

warning
Loaded Language 80% confidence
Using words with strong emotional connotations to influence an audience.

fact_checkFact-Check Results

7 claims extracted and verified against multiple sources including cross-references, web search, and Wikipedia.

info Single Source 3
help Insufficient Evidence 2
verified Verified 1
check_circle Corroborated 1
info
“Liu’s departure from Tencent after more than eight years”
SINGLE SOURCE
The provided evidence for this claim consists of general information about Tencent's headquarters, meeting software, and salary ranges. There is no mention of an individual named Liu or their departure from the company.
travel_explore
web search NEUTRAL — Tencent Meeting provides an easy-to-use, smooth, secure and reliable cloud-based high-definition video conferencing solution that enables customers and users to host or join meetings anytime, anywhere…
https://www.tencent.com/en-us/responsibility/combat-covid-19…
travel_explore
web search NEUTRAL — Software Engineer compensation in China at Tencent ranges from CN¥273K per year for T4 to CN¥2.2M per year for T12. The median yearly compensation in China package totals CN¥440K. View the base salary…
https://www.levels.fyi/companies/tencent/salaries/software-e…
travel_explore
web search NEUTRAL — Tencent is a Chinese multinational technology conglomerate and holding company headquartered in Shenzhen.Known as Tencent Helix, it will accommodate more than 23,000 employees across nearly 500,000 sq…
https://en.wikipedia.org/wiki/Tencent
info
“Hunyuan was introduced only a year earlier”
SINGLE SOURCE
The evidence provided consists of general Wikipedia and corporate descriptions of Tencent's founding and mission. There is no mention of the 'Hunyuan' AI model or its introduction date.
travel_explore
web search NEUTRAL — Tencent (Chinese: 腾讯; pinyin: Téngxùn) is a Chinese multinational technology conglomerate and holding company headquartered in Shenzhen. It is one of the highest grossing multimedia companies in the w…
https://en.wikipedia.org/wiki/Tencent
travel_explore
web search NEUTRAL — Mar 3, 2026 · Founded in 1998, Tencent is an Internet-based platform company using technology to enrich the lives of Internet users and assist the digital upgrade of enterprises. Our mission is "Value…
https://www.tencent.com/en-us/
travel_explore
web search NEUTRAL — Tencent, founded in 1998, is an internet company enhancing users' lives and supporting enterprise digitalization with its mission "Value for Users, Tech for Good".
https://www.tencent.com.cn/en-us/index.shtml
info
“OpenAI’s ChatGPT and Anthropic’s Claude Code [are examples of fanshi innovations]”
SINGLE SOURCE
While the evidence confirms that ChatGPT and Claude Code exist and are AI models/tools, none of the provided sources use the term 'fanshi' or categorize them as 'fanshi innovations'. This terminology appears to be specific to the original claim's source.
travel_explore
web search NEUTRAL — Claude is a series of large language models developed by Anthropic and first released in 2023. Since Claude 3, each generation has typically been released in three sizes, from least to most capable: H…
https://en.wikipedia.org/wiki/Claude_(language_model)
travel_explore
web search NEUTRAL — The AI revolution is here to stay, says Sam Altman, the CEO of OpenAI. In a probing, live conversation with head of TED Chris Anderson, Altman discusses the ...
https://www.youtube.com/watch?v=5MWT_doo68k
travel_explore
web search NEUTRAL — npm install -g @anthropic-ai/claude-code.Claude Code Router supports environment variable interpolation for secure API key management. You can reference environment variables in your config.json using…
https://github.com/musistudio/claude-code-router
help
“Chinese companies are either copying DeepSeek or US companies at the core technical level”
INSUFFICIENT EVIDENCE
The evidence directly contradicts the claim that Chinese companies are merely copying. One source explicitly states that DeepSeek 'has disrupted that notion' of copying and that its 'innovative techniques' have forced established players (including US companies) to 're-evaluate their approaches'.
travel_explore
web search NEUTRAL — Based in Hangzhou, Zhejiang, DeepSeek is owned and funded by High-Flyer, a Chinese hedge fund. DeepSeek was founded in July 2023 by Liang Wenfeng, the co-founder of High-Flyer, who also serves as the …
https://en.wikipedia.org/wiki/DeepSeek
travel_explore
web search NEUTRAL — But Chinese AI development firm DeepSeek has disrupted that notion. On Jan. 20, 2025, DeepSeek released its R1 LLM at a fraction of the cost that other vendors incurred in their own developments. Deep…
https://www.techtarget.com/whatis/feature/DeepSeek-explained…
travel_explore
web search NEUTRAL — DeepSeek's innovative techniques, cost-efficient solutions and optimization strategies have forced established players to re-evaluate their approaches.
https://www.forbes.com/sites/janakirammsv/2025/01/26/all-abo…
verified
“DeepSeek’s breakout moment early last year”
VERIFIED
The evidence confirms that DeepSeek released its foundational LLM series on November 29, 2023, and that this development drew significant interest and was described as a 'Sputnik moment' for AI, supporting the claim of a breakout emergence around that period.
travel_explore
web search NEUTRAL — Early Milestones and the Transition to Sparse Architectures. The release of the DeepSeek-LLM series on November 29, 2023, served as a foundational step, providing 7B and 67B parameter variants that de…
https://www.healthcare.digital/single-post/what-impact-has-d…
travel_explore
web search NEUTRAL — DeepSeek has also made significant progress on Multi-head Latent Attention (MLA) and Mixture-of-Experts, two technical designs that make DeepSeek models more cost-effective by requiring fewer computin…
https://www.wired.com/story/deepseek-china-model-ai/
travel_explore
web search NEUTRAL — This development has drawn significant interest in the tech world, especially after the company revealed its shockingly low operational costs. Marc Andreessen, a prominent tech investor, called DeepSe…
https://www.nigeriasun.com/news/274994070/deepseek-r1-one-of…
check_circle
“the launch of Anthropic’s Mythos model in April”
CORROBORATED
Multiple independent sources (The Conversation, AI News Roundup, and other web results) confirm that Anthropic launched the 'Mythos' model on April 7, 2026.
travel_explore
web search NEUTRAL — Mythos is one of Anthropic's latest models developed as part of its broader AI system called Claude. It encompasses the company's AI assistant and family of models, rivalling OpenAI's ChatGPT and Goog…
https://www.bbc.com/news/articles/crk1py1jgzko
travel_explore
web search NEUTRAL — While debuting Mythos, Anthropic said the model’s ability to find software flaws at scale could, if misused, pose serious risks to economies, public safety and national security. U.S. software stocks …
https://www.cp24.com/news/money/2026/04/20/what-do-we-know-a…
travel_explore
web search NEUTRAL — Claude Mythos Preview, the company’s latest general-purpose frontier model announced on April 7, triggered Anthropic’s ASL-4 safety classification — a tier defined as models with “high-uplift” capabil…
https://www.baus.ai/blog/ai-news-roundup-april-18-2026
+ 1 more evidence source
help
“domestic Chinese leader DeepSeek failed to reach the same heights it previously did with its latest V4 model”
INSUFFICIENT EVIDENCE
The evidence directly contradicts the claim that V4 failed to reach previous heights. The source 'DeepSeek V4 - 使用入口' describes V4 as having 'completely bid farewell to parrot-like repetition' and possessing 'extremely strong original thinking ability' in complex tasks, indicating an improvement over previous versions.
travel_explore
web search NEUTRAL — 强化学习赋能的推理引擎 DeepSeek V4 彻底告别了“鹦鹉学舌”，通过其独特的推理决策模块，能够自主进行多步逻辑推导，在处理数学难题、编程复杂架构和管理决策时表现出极强的原创思维能力。
https://deepseekv4.wiki/
travel_explore
web search NEUTRAL — 付费培训 DeepSeek爆火之后，中国大陆网络出现了很多针对DeepSeek在 电商 、 自媒体 、教育、 编程 等领域应用的培训课程，内容包括本地部署、提示语等，有些为免费，有些则须支付几十到上千元不等的费用。
https://zh.wikipedia.org/wiki/深度求索
travel_explore
web search NEUTRAL — 基于自研训练框架、自建智算集群和万卡算力等资源，深度求索团队仅用半年时间便已发布并开源多个百亿级参数大模型，如DeepSeek-LLM通用大语言模型、DeepSeek-Coder代码大模型，并在2024年1月率先开源国内首个MoE大模型（DeepSeek-MoE），各大模型在公开评测榜单及 ...
https://www.deepseek.com/
info Disclaimer: This analysis is generated by AI and should be used as a starting point for critical thinking, not as definitive truth. Claims are verified against publicly available sources. Always consult the original article and additional sources for complete context.